Data Analysis with R, Second Edition - Second Edition

Generalized Linear Model (GLM)

If you are surprised that linear regression can be generalized enough to accommodate classification, prepare to be astonished by generalized linear models!

GLMs are a generalization of regular linear regression that allow for other link functions to map from linear model output to the dependent variable, and other error distributions to describe the residuals. In logistic regression, the link function and error distribution is the logit and binomial respectively. In regular linear regression, the link function is the identity function (a function that returns its argument unchanged), and the error distribution is the normal distribution.

Besides regular linear regression and logistic regression, there are still ...